Social desirability bias is one of those hidden risks that can quietly undermine the quality of market research data. In essence, it’s when respondents provide answers they think are more socially acceptable or appropriate, even if those answers aren’t entirely true. This happens more often than you might think and can have a real impact on your research outcomes.
For instance, if you’re asking people how often they exercise, someone might “round up” their response to reflect the best week of their year, rather than their average activity. They’re not necessarily trying to mislead you, but there’s a subtle pressure to respond in a way that aligns with social norms or expectations. This is particularly common in sensitive areas like voting behavior, alcohol consumption, and personal habits (such as those related to eating, hygiene, and personal finance). It applies to B2B research too; in a survey on vendor evaluation practices among IT managers, respondents exaggerated the rigor of their processes, describing them as more structured and thorough than they actually were.
The good news is that there are practical ways to minimize the impact of social desirability bias, including:
- Use neutral wording and offer “prefer not to answer” options.
- Encourage honesty with scales and answer options that allow for very diverse responses (some people are reluctant to select an “extreme” response).
- Keep surveys anonymous and avoid situations where respondents may feel personally identified.
- Be extra careful when conducting phone, webcam, or in-person research. The lack of anonymity can unintentionally amplify this bias, as respondents may react to an interviewer’s facial expressions, body language, or tone of voice.
For a deeper dive, Research Rockstar students can access the job aid, “Tips to Avoid Common Bias Errors“, in the Questionnaire Design 201 course through the Training Portal.
The post How to Get Honest Answers: Combating Social Desirability Bias in Surveys first appeared on Research Rockstar Training & Staffing.
